The Shape of Inequality And Its Impact on Growth

July 28th, 2011 at 5:05 pm

The Iron Man (John Irons) makes an important point here: economic inequality doesn’t grow just because folks at the top do better.  It also grows because folks at the middle and bottom do worse.

This figure from EPI shows the fanning out of real hourly wages, indexed to 100 in 1973, over the last few decades.  Clearly, the top continuously pulls away from the pack, as can be seen by the lines for the 90th and 95th wage percentiles.

But the slide also shows ways in which the shape of inequality, at least re wages, has changed over these years.

Real Hourly Wages, by Wage Percentile, 1973-2009

There was little wage inequality in the 1970s, but in the 1980s, the bottom was falling, the middle flat, and the top rising—a basic fanning out of the wage distribution.

The latter 1990s were once again different.  The high end continued to rise but low and middle real wages grew together, and at a pretty decent clip.  The cause of the nice bump was the full employment conditions that prevailed for a few years back then, and as I’ve stressed, that dynamic leads to broad-based growth which pushes back against rising inequality.

Middle and low-wages flattened in the much weaker job market of the 2000s, while the top just kept on ticking. 

One other point on this—a hypothesis, really, and one that needs work but is likely important.  I pointed out above that faster growth and especially lower unemployment was inequality-reducing, but I believe there’s a feedback loop here.  Full unemployment delivers more broad-based gains, and this feeds back into longer, more durable, and just plain better recoveries.

I’d go further—still in hypothesis mode, but I’ll bet I’m right.  High levels of inequality depress longer-term growth by depressing more broad-based consumption—you end with a lot going on at Walmart and Nordstrom without enough going in the middle—and dampening investment both in human and physical capital.

There’s a bit of research on this point, and it’s a least mildly supportive of the above contentions, but more to come on this, I assure you.  If I’m right, it’s an important and underappreciated downside of high inequality.

Print Friendly, PDF & Email

13 comments in reply to "The Shape of Inequality And Its Impact on Growth"

  1. Geoffrey Freedman says:

    Another aspect of this is the growing disparity in net worth between the wealthiest and most people in this country. Since 2001 the top 20% has gone from having slightly more than 80% of the aggregate net worth of this country to about 85% of the agregate net worth of this counrty in 2010. This is a huge change in the distibution of wealth on a macro econmoic level is is devistating to agegate consumption. There are main two reasons for this redistribution in wealth, one of which is obvious, one of which is not so obvious.

    The obvious issue is that the main contributor to a middle income family’s worth is the home. With the fall in the value of housing, the net worth of middle income families has been devastated. The other issue that is not so obvious is the effect of the Bush tax cuts.

    Depending on how you frame this issue, the effect can be skewed to favor the middle class or the upper class. If you look at it the way Bush did, you focus on the percentage reductions of income tax liabilities of taxpayers at various points of the income distribution, the tax cuts look slightly progressive. If you look at the cuts this way, the implicit assumption is that a tax cut is distributionally neutral if it results in the same percentage reduction in income tax liability for taxpayers in all income classes. If you look at the tax cuts via this prism it appears that the income percentiles actually pay a greater share of taxes. This is the same way Reagan justified his tax cuts.

    However if you frame the tax cuts in the percentage of change in actual after tax income the results look quite different. If you frame in in the context of the actual amount of changes to after tax income, the results are even more startling. For example the top .1 percent of tax payers recieved an average income tax reduction of approximately 149,500. The top 20% received about 4370 in income tax reduction. The lowest 20% recieved about $17 in actual tax cuts. The dollar value at the lower ends were obviously used in consumption. The dollars at the high end we most likely not returned to the aggregate consumptive economy, the were most likely used in investment activity whcih does not increase GDP and decreases aggrigate consumption by effectively taking dollars out of the economy.

    So the argument that the upper percentile of tax payers pay a disproportionate share of income taxes requires that you look at taxes via the prism of the percentage reductions of income tax liabilities of taxpayers at various points of the income distribution. If you look at the changes in actual after tax income the results are very different, and the Bush taxe cuts were actually very regressive in nature and harmful to aggregate consumption.

    Another Republican con job?

  2. Kevin Rica says:

    In the last 10 years, the U.S. economy has create zero net new jobs. Yet we have accepted over 9 million legal immigrants alone. That doesn’t include illegal immigrants.

    The policy endorsed by the leadership of the Democratic Party and the Chamber of Commerce is that we must bring in more immigrants to prevent labor shortages (although they never explain where they think they see labor shortages). In graduate school and real life, labor shortages raise wages.


    How can you reconcile complaining about low wages and supporting a political party which endorses the Chamber of Commerce’s right to cheap labor?

    You know that these things are mutually exclusive.

  3. Tom says:

    I can’t speak for Jared, Kevin, but I can tell you how I reconcile those positions.

    First, the idea of deporting people who took huge risks and made great sacrifices to become integral parts of our communities strikes me as cruel and self-defeating. Therefore, I support paths to citizenship.

    Second, for the sake of both the labor market stability you mention and national security, I support effective border controls (though preferrably not something reminiscent of the Berlin Wall).

    So there you have it: Paths to citizenship and effective border controls. Sound familiar? Something you could support perhaps?

    • Kevin Rica says:


      That stuff doesn’t protect American workers.

      Why are illegal immigrants more important than American workers?

      Why should they get the jobs instead of American workers?

  4. Misaki says:

    Full employment also means people don’t feel as pressured to work extra for their employer. There are few complaints about not being paid enough for the amount of time worked, although there will always be people in a tight financial situation due to whatever discretionary or mandatory purchases. But people who are constantly working overtime for their employer simply do not have the time to enjoy any money that they earn, and so have no reason to ask for more money due to the lack of utility from it.

    If, on the other hand, “low consumer demand” meant that people worked LESS instead of MORE on average, then their marginal utility for money would be higher, leading to more pressure on wages and more income inequality, even when ignoring the iPod effect from signalling and barriers to entry for sellers for higher product qualities.

    The great illusion is that when a developed country is in an economic recession, an individual working harder with less pay is what will contribute to the country escaping the recession. This popular assumption, which results from both the economic structure, the naive suggestions of economists, and the pattern of information evaluation from previous experience in modern society, is simply not true. Working harder simply makes the recession worse; this is the entire reason recessions and the “business cycle” exists.

    • Misaki says:

      >instead of MORE on average, then their marginal utility for money would be higher, leading to more pressure on wages and more income inequality

      Should have been income equality… always hope that minor mistakes like this are obvious to whoever reads them, but can never be too sure.

      Someone recently had the “amazing” revelation that academic economists do not always have accurate models. Sadly he did not proceed from there to the conclusion that simply predicting that unemployment will rise from 8.8% to 9.2% using data prior to the June job report is not as useful as knowing how to prevent that from happening. This is the consequence when people measure their success by how correct and socially popular they are, instead of in terms of results. A disconnect between academic economists and the real world, indeed.

    • Misaki says:

      While Mr John Irons did have another entry on pay and productivity, the paper it linked to started off by examining the gap between these two but ended by ended by declaring that the way to increase wages was by encouraging conditions that lead to productivity and service quality levels. It would seem the reason for this conclusion was that “overseas labour is reducing the demand for unskilled work in the US”, despite also mentioning that found that 50% of young B.A. holders in the US have a job that does not require a college degree. (cited as Sum and McLaughlin, 2010 and presumably the same Sum as the author of this article)

      Anyway, it may be worth mentioning that the ‘Ipod effect’ may actually be intensified in the social and psychological conditions that persist during a (actual or perceived) recession with high unemployment; spending on expensive status symbols does not actually seem to be “wasting” either money or the raw materials and labour that go into a product, since the consumer understands that some amount of profit is extracted from the purchase and they are explicitly aware the signalling value of the product makes up some of its cost.

      It then becomes a way of “hoping that their money reaches someone who needs it without selfishly wasting any products”; in other words, a lack of consumption becomes a virtue exactly as described in the detailed explanation of how to fix unemployment, even while a larger amount of consumption is what would actually lead to a reduction of unemployment.

      This train of thought can be described as follows: “Other people are unemployed and suffering. I should therefore work harder in my job to help them.” …(it’s hard to understand thinking that is as illogical as this)… “Since other people can’t afford to purchase products since they’re unemployed, I should avoid purchasing those products too so they don’t feel bad. If I give away all my money by purchasing expensive iPods instead, this money will magically reach people who are unemployed and not selling expensive iPods, because Apple is a nice company and they know to do with this money since they help kittens and stuff.”

      And, after all, it’s true! The people who own Apple are giving away their money to help kittens! …But people still don’t have jobs.

      reposting comment… at least my success rate in closing tags is improving on average.

  5. genauer says:

    When you download the EPI data and calibrate your “100” levels to the average of the period ~ 1985 – 2000 (post Reagan changes and pre Bush II changes), about 80 % of the differences disappear. And it shows the huge pay raise across all deciles by 10 % between 1996 – 2002, which was not justified by a productivity increase relative to the rest of the World. This was then financed by increasing debt and fake wealth. Otherwise it just reflects the Reagan und Bush II tax cuts. Let the Bush II tax cuts end for all folks, and everything will be fine.

  6. pjr says:

    A couple of additional observations. If the figure is showing mean rather than median real wages, it actually understates the spread. Second, the spread within the top 5 percent is huge–the top 2, 1, and .1 percent have done far better. Third, declining wage discrimination against women, combined with women joining the workforce, are very good things but hides the worse growth (actually non-growth) in real wages experienced by men in the lower percentiles.

    Political factors–especially tax policies since the 1980 election, federal/state anti-union policies, and a lack of policy priority on reducing unemployment levels and raising wages from the bottom–have greatly reinforced economic factors that encouraged the “great divergence.” With weak unions and to date no institutional substitute for their role in politics, it is difficult to picture a reversal of trends.

    • Paul J says:

      Say, Genaur, how do you download the EPI data?

      A few questions. Are the “wages” mean or median wages? If they are mean wages, the wages of almost all people in the 80th percentile and above are being overstated by the huge increase in wages of the upper 1 percent. This is even truer for those at the 90th and 95th percentiles.

      Second, are they just “wages” or do the figures include things like capital gains?

      And an observation. Without the relative wage gains during the Clinton years, which the Bush years only managed to flatten rather than drive back down to Reagan-era levels, most people would be below 100 today.

      • genauer says:

        To download the EPI data, just go to the link, Jared already mentioned above:
        and push the Download Excel button (in the center upper third of the page). Very convenient, but in this case undermines her argument.

        She states “real hourly wages by wage percentile”, sooo, this should not include other income like capital gains, as percentiles it is not a mean, the median is defined as the 50th percentile and since there is no averaging, the data for the 90th or 95 percentile are not distorted by what happens in the higher percentiles (practically only the highest 1 %) This is all written there, for you and me to see, they also have some text there at, but the site crashed just right now, when I tried to look up another graph.

  7. genauer says:

    There is no “great divergence”

    Please take a look on the slightly modified original file from EPI.

    When you normalize the income levels to where they have been between 1985 – 2000, and to the median (50th percentile), everything is within a noise level of relative 3 %, and a slight adaption for the upper 10 % in response to the tax laws. At marginal / maximal tax rates of 70% prior to Reagan it made perfect sense to have a little higher part of the compensation package in not tax relevant “corporate perks” like generous company pensions, country club, low rate loans and similar things.

    The real income distribution is as it ever was, the same as in 1913, as it is in Germany over the same 100 years, just modified slightly to what passes the counter (mostly tax statistics) at what point in the chain.

    • genauer says:

      Just a very simple rule: the 10th percentile has a factor of 2.0 lower than the median (income), the upper 10th percentile a factor of 2.0 higher, interpolate exponentially. Done.

      The same in the last 100 years, and in Germany, where I also looked in closer Detail, over time, accounting for household size, somewhat different acocunting, especially for tax. Housing costs, tax deductions, other benefits like government child support, etc.

      I am in a very unique position for that. I am/was for personal decisions (relocation, am I really better of as a high income guy in the US ? What happens if taking an entrepreneurial risk turns out bad = low end of the distribution : – )
      actually very interested in what is really the case. And how might that be different in 20, 40 years down the road.

      I did/do not have an interest to bias for any position (high versus low, USA vs Germany). And the Quintessence of a lot of data mangling: Its all the same, within some noise level of relative 5%